Website security · Raksha guide

AI Visibility Checker: Access, Content and Citation Readiness

An AI visibility checker reviews whether public content is accessible, understandable and ready to reference. Raksha’s score is a heuristic assessment of page signals—not a live test showing that ChatGPT, Perplexity or Claude has read, ranked or cited your site.

By Raksha · Reviewed 16 November 2026

Six measures in plain language

Readiness indicators, not independent measurements of AI reach
MeasureWhat it describesWhat it cannot prove
Citation coverageReference and content signals that support citation readiness.The number of real AI answers citing your site.
Answer-engine optimisation (AEO)Clear questions, answers and machine-readable context.That an answer engine will choose your answer.
Generative-engine optimisation (GEO)Supported access, structure and source-quality indicators.Placement in a generated response.
Classic SEOPage titles, descriptions, canonical and discovery signals.A guaranteed search position.
Entity coverageClear identity and related organisation or content context.A complete external knowledge graph.
Content authorityAuthorship, references and other page-quality signals.Independent verification of every factual claim.

The indicators overlap and are combined according to Raksha’s scoring rules. They should guide review, not be read as percentages of AI services that can see your site.

Bot access first—but distinguish bot purposes

Training crawlers, search crawlers and user-triggered fetchers can have different policies. GPTBot is not the same as OAI-SearchBot or ChatGPT-User; Google-Extended is not Google Search’s indexing crawler. Review current provider documentation before changing broad rules.

Check robots.txt, authentication, firewall responses and actual access logs. A permissive robots file does not guarantee a successful fetch, and blocking one training crawler does not automatically remove your site from all AI answers. Never expose private paths just to improve a readiness score.

Make public content worth referencing

  1. Answer a specific question directly before adding detail.
  2. Use stable URLs, clear headings and descriptive page titles.
  3. Explain authorship, sources and dates where relevant.
  4. Keep business identity and factual statements consistent.
  5. Link to useful documentation and maintain accurate structured data.
  6. Add a concise public content map when it helps readers.
Check your llms.txt content map

What a score of 50 means

It means the observed page signals received that score under the checks that ran. It does not mean a 50% probability of citation, 50% crawler access or a security risk. A partial scan, inaccessible page or changed template can affect the result.

Review individual observations, address actual blockers and compare runs of the same scope. Web Health findings are kept separate from security findings so missing content metadata does not inflate vulnerability counts.

How to measure real AI visibility

Record dated, repeatable prompts and the service used, inspect whether a response actually cites your URL and track results over time. Answers can vary with region, account and model changes. Server logs can show requests but do not prove that a request led to a citation.

Does llms.txt guarantee citations?

No. Nor does ranking on Google guarantee selection in an AI answer. Treat content-readiness checks and real citation observations as separate evidence.

Sources and further reading

Consult current official documentation before changing a production system. Examples are illustrative, not live findings or universal configurations.

Start with a site you are authorised to test

A free passive check needs no card. Sign-in, email confirmation, ownership rules and usage limits apply. It is not an unrestricted penetration test.